TAIMS: Transparent Adaptation in Intelligent Multimodal HumanMachine Systems

1
TAIMSTransparent Adaptation in Intelligent
Multimodal Human-Machine Systems

• project startMay 2002

2
team
Karon MacLean, Lead Kathleen Akins Cristina
Conati Sidney Fels Dinesh Pai Ronald Rensink
UBC Computer Science SFU Philosophy UBC Computer
Science UBC Elect. Computer Eng. UBC Computer
Science UBC Psych, Comp. Sci
3
backgroundinterface transparency

• users efforts are entirely directed at a
  task,and not at the interfacethat supports the
  task.
• user and machine forma seamless system

4
interface transparency

• perceptual transparencyinformation is absorbed
  without conscious attention (key may be rapid
  sensory responses)
• attentional transparencysystem directs
  attention to whats important without effort or
  disruption
• task transparencyappropriate information is
  supplied as user needs it

5
objectives

• design and evaluate user interfaces that adapt
  conscious and nonconscious user feedback
• study the interplay between conscious and
  nonconscious processing
• create a state-of-the-art realtime platform to
  support 1 2

6
objective 1 (ultimate goal)design and evaluate
user interfaces that adapt conscious and
nonconscious user feedback

• Explore potential for lowering overall user
  effort by shifting user engagement from conscious
  to nonconscious modes.
• User models updated through low-latency biometric
  sensing and interpretation of user state and
  interaction context
• Rely on context-sensitive thresholds determined
  in 2)

7
objective 2study the interplay between
conscious and nonconscious multimodal processing

• Determine process characteristics in visual,
  auditory and haptic domains
• thresholds for transparency in each modality
• how modalities interact
• map intelligence of nonconscious processing

8
objective 3create a state-of-the-art realtime
platform

• Build a research tool to support 1 and 2
• fast local network cluster
• coordinates sensor display servers with data
  and user model servers
• high communication bandwidths
• supports variety of refresh rates, platforms

9
approach
developrealtimeplatform
developuser models prototypeapplications
conductpsychophysicsexperimentsin
multimodalperceptualprocessing
time
10
status

• just starting
• Ben Forsyth (M.Sc., CS)beginning work on
  realtime platform
• multimodal experiments being planned